Request an intervention

Operational notes Testing

EN 303 215 sets no power consumption limits: what actually goes into the declared figure

7 min read

A metal picture frame propped against a white wall, the white mount still in place and the plain backing board where the photograph should be, black and white photograph
The frame is there, and so is the mount. Inside, the packing board has never been taken out: what the edge promised was never put in.

Your specification says power consumption compliant with the limits in EN 303 215. The line went through, the supplier declared its figure, the evaluation panel put it in the table. Two things, then, before the next meeting. Those limits are not in the document: the word limit appears in it four times, and not one of the four sets a threshold. And the figure you were given does not include cooling, redundant power or batteries — a good share of what you will later see on the electricity bill.

Four occurrences of limit, and the fourth refers back to itself

The document is ETSI EN 303 215 V1.5.1 (2025-06), Environmental Engineering (EE); Measurement methods and limits for power consumption in broadband telecommunication networks equipment: twenty-five pages, downloadable free of charge from the ETSI file server.

A count across the whole text gives four occurrences of limit or limits (limited, limitation and limiting are other words). The first is in the title. The second sits in a note to the definition of low-power state: the traffic threshold below which the equipment changes state. The third is in clause 4.2, where the power-saving states allow the operator “to further limit the power consumption of the equipment”. The fourth is the one that matters, in clause 4.1, immediately below the formula PBBport = PBBeq / Nports: “PBBport is the power consumption per port in W of the broadband network equipment for which the limits are defined in the present document”.

Defined in the present document, it says. The table of contents says otherwise: definitions (3), definition of power consumption (4), measurement methods (5), four informative annexes A to D, and annex E with the change history. No limits clause, no table of thresholds, no normative annex. The only table that carries watts, B.2, sits in an annex that warns first: “NPC values are not actual requirements but informative only”.

This is no recent pruning: in V1.3.1 (2015-04) the word appears four times in the same positions and the structure is identical. The gap between the title and the content has been stable for ten years.

What the document actually promises

The scope promises no limits: “The present document defines the power consumption metrics, the methodology and the test conditions to measure the power consumption of broadband fixed telecommunication networks equipment”. And it narrows straight away: “The present document does not cover all possible configuration of equipment but only homogenous configurations”. A shelf with a mix of cards is already outside the method, before anyone has measured anything.

Outside the perimeter, too, is anything that is not DSLAM DSL, MSAN, PON OLT or point-to-point OLT: “Other access technologies may be included in further versions of the present document”. May be — which is to say, not yet. And the box in the subscriber’s home has nothing to do with it: “The end-user equipment is handled in other documents” — ETSI EN 301 575 for CPE, ETSI EN 303 423 for networked standby.

What goes into the figure, and what does not

Clause 5.1.3 lists what counts as part of the equipment, and whose “power consumption shall be taken into account”: the network termination with its links towards the core or backhaul, the line termination, the backplane, the cooling system (“fans drawer inside the broadband system”, the fan tray inside the unit) and the normally operational power-supply unit.

Clause 5.1.4 lists what does not, and whose “power consumption shall not be added”:

  • the external AC-DC rectifier;
  • the room or outdoor cabinet ventilation and air conditioning unit (“Room or outdoor Cabinet Ventilation and Air Conditioning Unit”);
  • the auxiliary or redundant power unit;
  • the battery;
  • for DSLAM equipment, additional external signal processing (DSM and MIMO) where it is not implemented inside the line termination.

Read that list again with the electricity bill of an exchange or a street cabinet in mind: these are precisely the items that the declared figure does not contain. Then a closing line, easy to miss: the additional functions of a line card — MELT, vectoring, test access, channel bonding — are to be disabled during the measurement, and where they cannot be, the manufacturer declares the extra power separately, because “Such extra budget will not be considered in PBBport”.

The measurement point follows from all this: clause 5.1.5 places it at Interface A of ETSI EN 300 132-2, the DC input of the shelf, downstream of the rectifier and the battery.

In which room, and with what traffic

In a laboratory, says clause 5.1.1, and it says so with figures: 25 °C ± 2 °C, relative humidity between 30 % and 75 %, −54.5 V ± 1.5 V for equipment nominally at −48 V. Traffic is fixed by clause 5.1.6: packets from 64 to 1 518 bytes flowing in both directions, at 80 % of the net activation rate for maximum power consumption and at 10 % for the typical figure — two values, not one. On PON, every port is closed on a 15 dB attenuator. On instrument calibration, clause 5.1.2 uses should, not shall: if you want it as an obligation, you write it yourself.

If your specification still cites V1.3.1

It often does, because ITU-T L.1310 still refers to the 2015 edition. Annex E, Change history, holds a single row: November 2024, V1.4.1, “Introduced new technologies”. The document history at the back shows why: between April 2015 and June 2025 there is no intermediate publication.

A line-by-line comparison says what changed. The scope moves from “GPON OLT” to “PON OLT”; four new normative references come in — XGS-PON (ITU-T G.9807.1), 50G-PON (G.9804.3), the 25GS-PON MSA v3.0 specification, coexistence (G.9805) — and clause 5.3.1 gains the matching configurations, multi-speed combo included; annexes D and E appear. The list of exclusions in clause 5.1.4 is identical, word for word, to the 2015 one. On the cover, the transposition dates: adoption on 30 April 2025, national publication and withdrawal of conflicting national standards both by 31 January 2026.

One detail, if your referral runs through L.1310: clause 9.2.2 there, headed GPON OLT network equipment, points to “clause 5.2 of [ETSI EN 303 215]”. But clause 5.2 of EN 303 215 is the method for DSLAM and MSAN equipment. The one for OLT equipment is clause 5.3.

What to write instead of that line

Five entries, half a page. Method and edition: EN 303 215 V1.5.1 (2025-06), clause 5.2 for DSLAM and MSAN, clause 5.3 for OLT. You set the threshold: the maximum watts per port you will accept, because the document has none. Two values, not one: at 80 % and at 10 % of the net activation rate, with the declared homogeneous configuration beside them. Site consumption kept separate: rectifier, cooling, redundancy and batteries, which clause 5.1.4 excludes and your bill does not. The report under clause 5.5: hardware and firmware, active ports, instruments and calibrations.

The point

Power consumption compliant with EN 303 215 is not a commitment to a value: it is a measurement method with no threshold, and a perimeter that stops at the DC input of the shelf. In the specifications and compliance work we deliver, that line arrives complete — edition, the right clause for the technology, a threshold in watts per port, the two loads, site consumption kept separate — and it becomes the running order of the check carried out at acceptance, with measurements and report handed over together with the work. On the other axis, consumption does not stop at acceptance: it shifts with load, with the season, with the cards added three years later. Meter readings, per-shelf current draw, room temperatures, traffic carried and as-built records of the equipment in the field sit on a single map of the network, where an AI recognises the shelf that has stopped being efficient before the purchasing office notices, and a team of ours steps in without subcontracting. That AI runs inside the client’s perimeter, on-premises or in a dedicated cloud in Italy, with CSIDIA, the other company in the group.

Are you writing the consumption line right now, or comparing two OLTs whose figures do not add up? Talk it through with an engineer: the site visit is free of charge, and the perimeter of the measurement is fixed before the tender.

Sources